Statistics formula sheet

This has mean n and variance n(1 ).

The Poisson distribution:

Summarising data
Sample mean:
x=

This has mean and variance .

n
1X
xi .
n
i=1

Continuous distributions

Sample variance:
s2x

x exp()
for x = 0, 1, 2, . . . .
x!

p(x) =

n
1 X
1
=
(xi x)2 =
n1
n1

n
X

i=1

x2i

nx

Distribution function:
.
F (y) = P (X y) =

i=1

f (x) dx.

Sample covariance:
g=

1
n1

n
X

Density function:

(xi x)(yi y) =

i=1

1
n1

n
X

xi yi nx y

d
F (x).
dx

f (x) =

i=1

Evaluating probabilities:

Sample correlation:
g
r=
.
sx sy

P (a < X b) =

f (x) dx = F (b) F (a).

Expected value:

Probability

E(X) = =

Addition law:

xf (x) dx.

P (A B) = P (A) + P (B) P (A B).

Multiplication law:

Variance:
Var(X) =

(x )2 f (x) dx =

P (A B) = P (A)P (B|A) = P (B)P (A|B).

x2 f (x) dx 2 .

Hazard function:
Partition law: For a partition B1 , B2 , . . . , Bk
P (A) =

k
X

P (A Bi ) =

i=1

k
X

h(t) =
P (A|Bi )P (Bi ).

i=1

Bayes formula:

Normal density with mean and variance 2 :

1

P (A|Bi )P (Bi )
P (A|Bi )P (Bi )
P (Bi |A) =
= Pk
.
P (A)
P (A|Bi )P (Bi )
i=1

f (t)
.
1 F (t)

f (x) =
exp
2 2

2 

for x [, ].

Weibull density:
f (t) = t1 exp(t ) for t 0.

Discrete distributions

Exponential density:
Mean value:
E(X) = =

f (t) = exp(t) for t 0.

xi p(xi ).

xi S

Variance:
Var(X) =

(xi )2 p(xi ) =

xi S

This has mean 1 and variance 2 .

Test for population mean

x2i p(xi ) 2 .

xi S

Data: Single sample of measurements x1 , . . . , xn .

Hypothesis: H : = 0 .

The binomial distribution:

p(x) =

 

n x
(1 )nx for x = 0, 1, . . . , n.
x

Method:

Calculate x, s2 , and t = |x 0 | n/s.

Obtain critical value from t-tables, df = n 1.

 Reject H at the 100p% level of significance if |t| > c,

where c is the tabulated value corresponding to column p.

Calculate
s2 = (n 1)s2x + (m 1)s2y /(n + m 2).

Look in t-tables, df = n + m 2, column p. Let the

tabulated value be c say.

Paired sample t-test

Data: Single sample of n measurements x1 , . . . , xn which
are the pairwise differences between the two original sets
of measurements.

100(1 p)% confidence interval for the difference in

population means i.e. x y , is
(x y) c

Hypothesis: H : = 0.

(r

s2

1
1
+
n
m

Method:

Calculate x, s2 and t = x n/s.

Obtain critical value from t-tables, df = n 1.

Regression and correlation

Reject H at the 100p% level of significance if |t| > c,

where c is the tabulated value corresponding to column p.

The linear regression model:

yi = + xi + zi .
Least squares estimates of and :

Two sample t-test

Data: Two separate samples of measurements x1 , . . . , xn
and y1 , . . . , ym .
Hypothesis: H : x = y .

Pn

xi yi n x y
,
(n 1)s2x

i=1

and
= y x.

Confidence interval for

Method:
Calculate x, s2x , y, and s2y .

Calculate as given previously.

Calculate s2 = s2y 2 s2x .

Calculate

s2 = (n 1)s2x + (m 1)sy /(n + m 2).

xy
.
1
1
s2
+
n
m

Calculate t = r

Obtain critical value from t-tables, df = n + m 2.

Reject H at the 100p% level of significance if |t| > c,
where c is the tabulated value corresponding to column p.

=
Calculate SE()

s2
.
(n 2)s2x

Look in t-tables, df = n 2, column p. Let the

tabulated value be c.

100(1 p)% confidence interval for is c SE().

Test for = 0
Hypothesis: H : = 0.

CI for population mean

Data: Sample of measurements x1 , . . . , xn .
Method:
Calculate x, s2x .
Look in t-tables, df = n 1, column p. Let the
tabulated value be c say.

100(1 p)% confidence interval for is x csx / n.

Calculate
t=r

n2
1 r2

1/2

Obtain critical value from t-tables, df = n 2.

Reject H at 100p% level of significance if |t| > c,
where c is the tabulated value corresponding to column p.

Approximate CI for proportion

CI for difference in population means
Data: Separate samples x1 , . . . , xn and y1 , . . . , ym .
Method:
Calculate x, s2x , y, s2y .

p 1.96

p(1 p)
n1

where p is the observed proportion in the sample.

Test for a proportion

Hypothesis: H : = 0 .
p 0
Test statistic z = q
.
0 (1 0 )
n
Obtain critical value from normal tables.

Comparison of proportions
Hypothesis: H : 1 = 2 .
Calculate
p=

n1 p1 + n2 p2
.
n1 + n2

Calculate
p1 p2

z= r

p(1 p) n11 + n12

Obtain appropriate critical value from normal tables.

Goodness of fit
Test statistic
2 =

m
X
(oi ei )2
i=1

ei

where m is the number of categories.

Hypothesis H : F = F0 .
Calculate the expected class frequencies under F0 .
Calculate the 2 test statistic given above.
Determine the degrees of freedom, say.
Obtain critical value from 2 tables, df = .
Reject H : F = F0 at the 100p% level of significance
if 2 > c where c is the tabulated critical value.